blacklabelops / logrotate Goto Github PK

We'd love to be able to use a pinned version of logrotate - mostly so we don't run latest - and a new release would be greatly appreciated :)

some apps need a signal before the close the file

how do we do that with this?

I'm using this in my docker-compose.yml:

 logrotate:                                                                                                                                                        
    image: "blacklabelops/logrotate"                                                                                                                                
    volumes:                                                                                                                                                        
      - /var/lib/docker/containers:/var/lib/docker/containers                                                                                                       
    environment:                                                                                                                                                    
      - LOGS_DIRECTORIES=/var/lib/docker/containers                                                                                                                 
      - LOGROTATE_INTERVAL=hourly                                                                                                                                   
      - LOGROTATE_COPIES=7                                                                                                                                          
      - LOGROTATE_SIZE=10M                                                                                                                                          
      - LOGROTATE_COMPRESSION=compress    

I don't see any compression when it rotates.
The logs were also much bigger when I first ran it and it didn't split it up in files of max 10M size, it just moved the old file to .1

executing: /bin/bash -c /usr/bin/logrotate.d/update-logrotate.sh; /usr/sbin/logrotate --state=/logrotate-status/logrotate.status /usr/bin/logrotate.d/logrotate.conf
find: "/var/log/docker: No such file or directory
find: /var/log/nginx": No such file or directory

├── docker-compose.yml
└── logs
    ├── file.log
    └── nginx
        ├── access.log
        └── error.log

This is my docker-compose.yaml setup

---
version: '2.4'
services:
  ssl:
    image: nginx:latest
    ports:
      - 2020:80
    volumes:
      - ./logs/nginx:/var/log/nginx
  logrotate:
    image: blacklabelops/logrotate:1.3
    environment:
      - LOGS_DIRECTORIES="/var/log/docker /var/log/nginx"
      - LOGROTATE_INTERVAL=hourly
      - LOGROTATE_COMPRESSION=compress
      - LOGROTATE_COPIES=30
      - TZ=Asia/Singapore
      - LOGROTATE_CRONSCHEDULE=* * * * * *
    volumes:
      - ./logs:/var/log/docker
      - ./logs/nginx:/var/log/nginx

Any way to log rotate a subfolder of a volume?
I've tried:

volumes:
  - /var/lib/docker/volumes/wp_docker_litespeed_lsws_sites/_data/localhost/wp_logs:/var/log/wp

and

LOGS_DIRECTORIES: "/var/lib/docker/containers /var/log/docker /var/lib/docker/volumes /var/log/wp"

I want to rotate log files right now with -e "LOGROTATE_PARAMETERS=f". But it's inoperative.

Hi is there some way to do a healthcheck when using blacklabelops/logrotate?

The default logrotate_croninterval="1 0 0 0 * *" (at

2016/11/21 16:40:23 Beginning of range (0) below minimum (1): 0

The 4th (day of month) can not be 0, it must be in range 1-31.

I am unsure what the intended value is, but I think it must be at least hourly looking at the lowest possible logrotate interval. If this is the case a cron interval of 0 0 * * * * would work. (every hour at 0 minutes and 0 seconds).

I am working around this by providing a valid LOGROTATE_CRONSCHEDULE as environment var.

How would one go about sending a signal from one container process to another container process?

For nginx, the worker needs to be gracefully restarted to remove the file handle on the log files via something like:

postrotate
  [ ! -f /var/run/nginx.pid ] || kill -USR1 `cat /var/run/nginx.pid`
endscript

But, as a sidecar, how would you do this without exposing the docker control socket?

-L option is missing in find invocation in that script.

Hi,
I am trying to deploy log rotate in the K8s cluster as a sidecar container but getting permission denied error.
touch: /usr/bin/logrotate.d/logrotate.conf: Permission denied

Is it possible to run logrotate w/o root privileges?

I would like to setup this cool container to rotate syslog file, but it looks like it cannot be done with the envs provided, since the code assume you have files with extension, or am I missing something?

Support for logrotate parameters maxage and minsize.

Hi, one quick question, if a container dies/recreated with a new container ID, and the old one is deleted, e.g. by K8s, will the rotated logs expire after the expected date?

I guess not, since the config is updated to new containers. If this is true, is there a way to handle this?

Thanks!

I will stop supporting and implementing Docker images starting in early 2019.

Repositories will be transferred to blacklabelops-legacy and archived for read only purpose.

The following repositories and images are affected:

January 2019

• blacklabelops/crowd
• blacklabelops/nginx
• blacklabelops/bitbucket
• blacklabelops/java
• blacklabelops/letsencrypt
• blacklabelops/postgres

March 2019

• blacklabelops/baseimages
• blacklabelops/confluence
• blacklabelops/jira
• blacklabelops/volumerize
• blacklabelops/logrotate

I have many log files present in /var/log/company_name to be rotated on an hourly basis into /var/log/company_name/rotated, keeping the past 24-hours. All logs seem to be rotating except a select few, which grow to gigabytes in size.

Upon inspecting the logrotate logs with docker logs logrotate, it appears that all the logfiles that aren't rotating have the following message printed when processing:

Found new file /var/log/company_name/mysql_error.log, Processing...
File has unknown user or group: , user: UNKNOWN, group: video

However, logfiles that are properly rotated have the following message printed:

Found new file /var/log/company_name/wkhtmltopdf.gunicorn.log, Processing...
Inserting new /var/log/company_name/wkhtmltopdf.gunicorn.log to /usr/bin/logrotate.d/logrotate.conf

Other notes:
User: root
Logrotate version: 1.2
Logrotate hash: sha256:e0d5e50a8716d7859079df4849f9034c56d0cb38ed46ad118b87e41f8a9437d6
Capabilities: dac_override, setgid, setuid
Network: none
Environment:

LOGROTATE_INTERVAL=hourly
LOGS_DIRECTORIES=/var/log/company_name
LOGROTATE_OLDDIR=/var/log/company_name/rotated
LOGROTATE_LOGFILE=/var/log/company_name/logrotate.log
LOGROTATE_COPIES=24
LOGROTATE_COMPRESSION=compress
LOGROTATE_STATUSFILE=/company_name/logrotate_status/logrotate.status
LOGROTATE_PARAMETERS=vf

I also tried enabling all capabilities and the network without any luck.

The error message appear accurate when running an ls -lahS on the log directory. But I don't understand why logrotate won't touch those files or how to force it to work with those files.

env:

    - name: LOGS_DIRECTORIES
      value: "/var/lib/docker/containers /var/log/docker"
    - name: LOGROTATE_INTERVAL
      value: "hourly"
    - name: LOGROTATE_OLDDIR
      value: "/logs"
    - name: LOGROTATE_SIZE
      value: 2k

`

Hello, we would like to use * to match all potential candidates, the reason is our service might generate logs with timestamp as suffix. Thank you.

I have the fallowing docker compose that is launched within a docker swarm. The rotation interval is set to hourly, but the logs are rotated daily.

version: '3.3'

services:
  log-roate:
    image: blacklabelops/logrotate
    environment:
      - LOGS_DIRECTORIES=/var/lib/docker/containers
      - LOGROTATE_COPIES=100
      - LOGROTATE_COMPRESSION=compress
      - LOGROTATE_INTERVAL=hourly
      - LOGROTATE_DATEFORMAT=-%Y-%m-%d-%s
    volumes:
      - type: bind
        source: /var/lib/docker/containers
        target: /var/lib/docker/containers
    deploy:
      mode: global

configurations as below:

• name: LOGS_DIRECTORIES
  value: /var/log/cmcapi
• name: LOGROTATE_COPIES
  value: "3"
• name: LOGROTATE_SIZE
  value: 100M
• name: LOGROTATE_MAXAGE
  value: "2"
• name: LOGROTATE_CRONSCHEDULE
  value: '0 */1 * * *'

but the LOGROTATE_MAXAGE does not work. there still are 3 days logs.

Hello,

I'm trying to use the logrotate docker image. I am overriding the LOGROTATE_CRONSCHEDULE environment variable. However, it is not having an impact on the logrotate command. Can you take a look?

environment:
        - "LOGS_DIRECTORIES=/home/logs"
        - "LOGROTATE_CRONSCHEDULE=* * * * * *"
        - "LOGROTATE_LOGFILE=/tmp/rotate-log"
        - "LGROTATE_PARAMETERS=vf"
        - "LOGROTATE_DATEFORMAT=-%Y-%m-%d-%s"
        - "LOGROTATE_COPIES=5"

Hi

I'm not sure if this container is abandonware or not, but I just discovered and implemented it, and it seems perfect for me with one small exception.

I have all of my containers set to place all of their logs in a central location on the host (/mnt/e/logs) and then a subdir per container.

In my case, it would be cleaner and more efficient to just mount the top-level dir as the volume mount, and then use an env var to ignore specific dirs (some containers have built in rotate mechanisms that I don't need this container to rotate for them).

Currently, I have to mount something like 20 containers, and also add them each on the single env var line, like this:

LOGS_DIRECTORIES: '/mnt/e/Docker/Logs/authelia /mnt/e/Docker/Logs/mysql /mnt/e/Docker/Logs/something /mnt/e/Docker/Logs/somethingelse' (etc., for another 15 or so more)

Having an env var something like:

LOGS_IGNORE_DIRECTORIES: '/mnt/e/Docker/Logs/ignorethisone /mnt/e/Docker/Logs/ignorethisonetoo'

Would be much nicer, because then I'd just mount the top-level dir, let logrotate recurse through all subdirs, and ignore just like, 3 directories. :)

Standard init linux error showed up. I updated entry point from #!/bin/bash to #!/bin/sh as it works for many people but it didn't work for me.
Any suggestion to solve it ?

Thanks for a great project. I have only a small (yet getting bigger) problem:

$ ls -lh
-rwxrw-rw-    1 root     root        1.6K Dec  3 01:14 alternatives.log
drwxrw-rw-    2 root     root        4.0K Dec  3 01:42 apt
drwxr-xr-x    2 root     root        4.0K Nov 19 03:59 docker
-rwxrw-rw-    1 root     root      527.4K Dec  9 05:15 docker.log
-rwxrw-rw-    1 root     root       36.5M Dec  4 00:00 docker.log-20171204.gz
-rwxrw-rw-    1 root     root        1.2M Dec  8 00:00 docker.log-20171208
-rwxrw-rw-    1 root     root       59.4K Dec  3 01:42 dpkg.log
-rw-r--r--    1 root     root         510 Dec  3 01:42 fontconfig.log
-rwxrw-rw-    1 root     root      314.0K Dec  9 05:15 lastlog
-rwxrw-rw-    1 root     root        1.7G Dec  9 05:17 messages
-rwxrw-rw-    1 root     root      663.5M Dec  9 05:17 secure
-rwxrw-rw-    1 root     root        2.3G Dec  9 05:17 syslog

As you can see secure, messages and syslog weren't touched by logrotate. Is it possible for you guys to have an option where we can add LOGS_FILES?

The new image is Alpine Docker 1.9 conform. Dockerhub is still on Docker 1.8 build slaves. This ticket is to remind me to check builds.

Imagelayers also seems to have problems when Dockerhub builds fail.